Professional Documents
Culture Documents
ISSN: 0122-5383
ctyf@ecopetrol.com.co
ECOPETROL S.A.
Colombia
López, Diana-C.; Mahecha, Cesar-A.; Hoyos, Luis-J.; Acevedo, Leonardo; Villamizar, Jaime-F.
Optimization model of a system of crude oil distillation units with heat integration and metamodeling
CT&F Ciencia, Tecnología y Futuro, vol. 3, núm. 5, diciembre, 2009, pp. 159-173
ECOPETROL S.A.
Bucaramanga, Colombia
How to cite
Complete issue
Scientific Information System
More information about this article Network of Scientific Journals from Latin America, the Caribbean, Spain and Portugal
Journal's homepage in redalyc.org Non-profit academic project, developed under the open access initiative
OPTIMIZATION MODEL OF A SYSTEM OF CRUDE OIL DISTILLATION UNITS WITH HEAT INTEGRATION AND METAMODELING
OPTIMIZATION MODEL
OF A SYSTEM OF CRUDE OIL
DISTILLATION UNITS WITH
HEAT INTEGRATION AND
METAMODELING
Diana-C. López1, Cesar-A. Mahecha2*, Luis-J. Hoyos2, Leonardo Acevedo3 and Jaime-F. Villamizar3
1
Convenio Ecopetrol S.A. - Instituto Colombiano del Petróleo - Universidad Industrial de Santander (UIS), Bucaramanga,
Santander, Colombia
2
Ecopetrol S.A. - Instituto Colombiano del Petróleo, A.A. 4185 Bucaramanga, Santander, Colombia
3
Universidad Industrial de Santander (UIS), Escuela de Ingeniería Química, Bucaramanga, Santander, Colombia
T he process of crude distillation impacts the economy of any refinery in a considerable manner. Therefore, it is
necessary to improve it taking good advantage of the available infrastructure, generating products that conform
to the specifications without violating the equipment operating constraints or plant restrictions at industrial units.
The objective of this paper is to present the development of an optimization model for a Crude Distillation Unit (CDU)
system at a ECOPETROL S.A. refinery in Barrancabermeja, involving the typical restrictions (flow according to pipeline
capacity, pumps, distillation columns, etc) and a restriction that has not been included in bibliographic reports for this
type of models: the heat integration of streams from Atmospheric Distillation Towers (ADTs) and Vacuum Distillation
Towers (VDT) with the heat exchanger networks for crude pre-heating. On the other hand, ADTs were modeled with
Metamodels in function of column temperatures and pressures, pumparounds flows and return temperatures, stripping
steam flows, Jet EBP ASTM D-86 and Diesel EBP ASTM D-86. Pre-heating trains were modeled with mass and energy
balances, and design equation of each heat exchanger. The optimization model is NLP, maximizing the system profit.
This model was implemented in GAMSide 22,2 using the CONOPT solver and it found new operating points with bet-
ter economic results than those obtained with the normal operation in the real plants. It predicted optimum operation
conditions of 3 ADTs for constant composition crude and calculated the yields and properties of atmospheric products,
additional to temperatures and duties of 27 Crude Oil exchangers.
Keywords: crude distillation units system, CDU, optimization, NLP, metamodels, operational constraints, energy constraints
(restrictions), heat exchanger network.
L
a destilación de crudo es un proceso que impacta considerablemente la economía de cualquier
refinería y por tanto es necesario mejorarlo aprovechando la infraestructura instalada, generando
productos dentro de especificaciones, sin violar las diferentes ventanas operativas o restricciones
de planta que existen en las unidades industriales. El objetivo de este artículo es presentar el desarrollo de
un modelo de optimización de un Sistema de Unidades de Destilación de Crudo (UDCs) perteneciente a
la refinería de Barrancabermeja de ECOPETROL S.A. que involucra las restricciones de planta típicas (flujo
por capacidad de tuberías, bombas, torres de destilación, etc.) y una restricción aún no contemplada en la
literatura para modelos de este tipo: la integración energética de las corrientes de las Torres de Destilación
Atmosféricas (TDAs) y las Torres de Destilación al Vacío (TDVs) con la red de intercambiadores de calor
que precalientan el crudo. Por otro lado, las TDAs fueron modeladas a través de Metamodelos en función
de las temperaturas y presiones de las torres, los flujos y temperaturas de retorno de los pumparounds, los
flujos de vapor de despojo, el PFE de la ASTM D-86 para el Jet y el Diesel. Los trenes de precalentamiento
fueron modelados con balances de masa y energía, y la ecuación de diseño de cada intercambiador. El
modelo de optimización es de tipo NLP, maximizando la utilidad del sistema. Este modelo se implementó
en GAMSide 22.2 usando el solver CONOPT y predice nuevos puntos de operación óptimos con mejores
resultados económicos que aquellos obtenidos con la operación normal en las plantas reales. El modelo
calculó las condiciones de operación óptimas de 3 TDAs para un crudo de composición constante y calculó
los rendimientos y propiedades de los productos atmosféricos, adicionalmente a las temperaturas y duties
de 27 intercambiadores de crudo.
Palabras Clave: sistema de unidades de destilación de crudo, UDC optimización, PNL, metamodelos, restricciones
operacionales, restricción energética, red de intercambiadores de calor.
mentation, execution, and convergence, surpassing the CDU1 and CDU2 share the network NET1 con-
rigorous models (Palmer & Realff, 2002a and 2002b). formed by 17 exchangers and the NET2 is composed
Crude oil preheating trains were represented by rigo- of 10 heat exchangers. The vacuum residue stream from
rous exchanger models. Metamodels were developed CDU3 interacts with CDU2 through the NET1 network
from previous PRO/II rigorous models of the industrial preheating the Crude Oil that feeds this unit.
units. Metamodels of the distillation towers and rigorous
models of the heat exchange trains were integrated in PRO/II Model Construction
the GAMSide 22.2 - based optimization algorithm that A rigorous model using the commercial simulator
maximizes the CDU system profits, finding optimum PRO/II ® version 8.0 was constructed for each CDU.
operation conditions for each atmospheric distillation As thermodynamic packages, the Grayson-Streed and
tower. Braun K10 (BK10) equations of state were used in the
atmospheric and vacuum towers respectively. Feedstock
stream was characterized using the TBP curve, API grav-
ity curves, sulfur content, neutralization number (NN),
METHODOLOGY Conradson Carbon content (CCR) and lightends content
(C1, C2, C3, iC4, nC4, iC5, nC5, hexane, H2O, CO2, CO,
The system to optimize comprises three crude oil N2 and H2). PRO/II models were tuned with the infor-
mation from industrial runs. Regarding ADTs, the set of
distillation units (CDU1, CDU2 and CDU3) and these
manipulated variables is conformed by the ope-ration
crude units are composed by three atmospheric distilla-
variables (Crude flow, input and output furnace tempera-
tion towers with different design characteristics (ADT1,
tures, temperature and pressure on the top of the column,
ADT2 and ADT3), three atmospheric furnaces (AF1,
pumparounds flows and return temperatures, stripping
AF2 and AF3), two vacuum towers (VDT1 and VDT2) steam flows, condenser temperature and pressure). A
and two heat exchanger networks (NET1 and NET2). quality control variable of the JET and DIESEL products
Figure 1 shows a simplified diagram of the CDUs studied is also included: End Boiling Point (EBP) ASTM D-86
in this work. distillation curve.
CDU 1 CDU 3
ADT 3
NET1 ADT 1
NET2
AF1
VDT 1
AF 3
VDT 2
ADT2
CRUDO OIL
CDU 2 NAPHTHA AGO
JET REDC
DIESEL VACUUM
AF 2 RESIDUE
Construction of Atmospheric Tower Metamodels The input variables set is composed for the PRO/II
For the purpose of this research, non-linear semi- manipulated variables and changes depending on the
rigorous models (Metamodels) of each atmospheric design of the tower (see Table 1).
tower were constructed. The methodology used to
develop the Metamodels includes:1) Selection of input The data collection plan or Design of Experiments
variables, 2) Design of Experiments (Latin Hypercube), (DoE) used was the sampling process by Latin Hypercube
3) Generation of data using PRO/II models, 4) Parame- (LHS: Latin Hypercube Sample) which provides unique
trization of Metamodels using multivariate stepwise values for each point and exhibits better dispersion than
regression, 5) Validation and evaluation of their sig- other sampling procedures such as the random and grid
nificance. Output variables in Metamodels were flows, sampling (Palmer & Realff, 2002a and 2002b).
temperatures, and properties of all the streams leaving
the towers (products and pumparounds toward NET1
The Metamodels were a second-order polynomial
and NET2). The properties calculated by Metamodels
function with binary interaction:
were API gravity, ASTM D-86 curve (10, 30, 50, 70
and 90%) and calorific capacities. Pumparound duty
Metamodels of ADT1, ADT2 and ADT3 were created,
which were linked to the NET1 and NET2.
Where ŷ(X) was the output variable vector (flows, analysis were used for the detection outliers within the
temperatures, properties, etc.), X is the normalized input Metamodel construction set. In order to measure the
variable vector, (feed flow, operating conditions, and Metamodel prediction capability, the R2Prediction (Mont-
quality properties) with i=1,..,13 for ADT1, i=1,..,138, gomery, 2001) was calculated. Each metamodel was
for ADT2 and i=1,..,17 for ADT3. Each output varia- evaluated by a validation set with 1.000 simulated points
ble has a Metamodel with characteristic parameters in PRO/II to verify their prediction quality before using
according to its behavior in regards to the input varia- them in the optimization model. Figure 2a shows the
bles. For ADT1, 105 parameters were calculated; for correlations between calculated volumetric flow for the
ADT2, 190 parameters and for ADT3, 171 parameters ADT2 products with Metamodels and the results found
were found. in PRO/II. Figure 2b shows the error distribution of
product flow Metamodels for ADT1 and ADT2.
Furthermore, 5000 sampling points were used in
Metamodels construction (N=5000). These points Atmospheric tower Metamodels showed a maxi-
represented 5000 simulation runs for each distillation mum average error of 2,5% with a standard deviation
tower (ADT1, ADT2 and ADT3) using the configured of 0,41% for flows, while for the rest of the properties
and tuned rigorous models in PRO/II. Due to the random the maximum average error was 2,0% with standard
nature of LHS for each independent variable and to deviation of 0,75%.
the complexity of the distillation process, 100% con-
vergence of the model is not attained. Therefore, only Construction of Heat Exchanger Integrated Networks
converged data was used in Metamodels construction.
NET1 and NET2 of the CDU system were modeled
through energy balances by Equation 2 on the hot side,
Parameters for each Metamodel were estimated by Equation 3 on the cold side and by the design equa-
based on the matrix of experimentation results in PRO/ tion expressed in Equation 4.
II, using multivariate stepwise regression. During re-
gression, input variables were normalized between -1
and 1, and output variable results (ŷ(x) were in the real
magnitude corresponding to each variable.
35.000
DIESEL AGO GOA REDC 5
30.000
y = 0,9659x + 271,34 y = 1,0021x - 1,8966 y = 1,0016x - 33,274
NAPHTA-ADT1 JET-ADT1 DIESEL-ADT1
25.000 R2 = 0,967 R2 = 0,9996 R 4
NAPHTA-ADT2 JET-ADT2 DIESEL-ADT2
Metamodels
20.000
3
Error (%)
15.000
10.000 NAPHTA
NAFTA JETJET 2
y = 1,0015x - 9,7902 y = 0,9714x + 160
5.000 1
R2 = 0,9999 R2 = 0,9623
0
0 5.000 10.000 15.000 20.000 25.000 30.000 35.000 0
NAPHTA-ADT2 JET-ADT2 DIESEL-ADT2 0 150 300 450 600
PRO/II AGO-ADT2 REDC-ADT2 Identity Experiments
a) b)
Figure 2. Flow Metamodel (bl/d) in a validation set: a) Metamodels and PRO/II Correlations. b) Errors (%)
Table 2. Flow Plant Restrictions of Products Pumping and Furnace Duties Plant Restrictions
The path flow of each fluid through the whole set of -REDC) from all the atmospheric towers Ua (ADT1,
exchangers was considered in network configuration, ADT2 and ADT3). The raw material cost is defined for
including recycles and bypass from real plants. all crudes entering to each atmospheric tower Ua. The
operation cost is defined considering all the resources
Determination of CDUs Operating Constraints consumed in the process, R, such as fuel, compressed air,
There are different operating constraints within an fresh water, industrial water, steam and electricity.
Table 3. Plant Restrictions of Crude Oil Preheating Exchangers: Temperatures and Areas
Table 4. Feed and Product Prices of an Table 5. Operation Cost of Crude Distillation Units (US$/bl of feedstock)
Atmospheric Tower (US$/bl)
STREAM PRICE
CRUDE 36,2
NAPHTHA 76,6
JET 91,5
DIESEL 89,6
AGO 20,0
REDC 15,0
Source : Refinery Global Model 2001 -ECOPETROL S.A.
solved with Non-Linear Programming algorithms (NLP). model that generates the operational and quality condi-
This research used the CONOPT solver since it correctly tions called the Optimum Feasible Point.
manages correctly models with many non-linear restric-
tions and a moderate number of degrees of freedom. These The algorithm in Figure 3 starts from available runs plants
are characteristics of the problem analyzed here. The in CDUs and/or converged points of individual models
Initial Estimation (IE) was a set of operating conditions in PRO/II of ADT1, ADT2 and ADT3 such as IE, which
close to an operation of the industrial unit. Therefore, an enters the CDU metamodel optimizer in GAMS. Next
operation with low yield of REDC and maximum Jet or the run is completed and the Optimum Variables (Vopt) are
Diesel production were selected. obtained; and they are compared to the Initial Estimation
Variables (VIE). The cycle finishes when these values are
Running the Optimization Model and Results close enough to each other (<TOL).
Once the objective function and the model restrictions
were configured, and the solver in GAMS was specified, Optimum Point Sensitivity Analysis
the next step was to solve the optimization problem. The Due to the fact that it is a NLP optimization, the new
results of the model included the profit of the system and operational point found by GAMS is a local optimum
of each CDU. Regarding ADT1, ADT2 y ADT3, they point. Several IE were tested in order to figure out the
included: operating variables, Crude flow, products yields, contour of the objective function. The optimum points
flow, temperatures and properties of both products and were compared and the best one was selected as the most
pumparounds. The following is included for NET1 and profitable of the system.
NET2: Duties, temperatures and properties of Crude and
hot streams at the entry and exit of each exchanger. The
furnace duty was calculated.
Q
Qpa,t
hx Hxpa
hx,u
(20)
AGO 3,8% 2,4% 1,8% 2,1% 1,8% 1,8% 4,0% 1,8% 1,0% 3,9% 1,8% 1,1%
REDC 49,8% 53,2% 58,4% 49,1% 49,9% 58,4% 46,1% 47,8% 44,8% 46,4% 48,0% 43,2%
exchange within and between CDUs, and the initial es- of 4,4%. The optimum results from EI-1 of Tables 6-8
timation, the algorithm could predict a new operational represent an increment in the system profit of about two
point with better economic results than those obtained hundred million dollars (M US$ 200) per year without
with the normal operation. The summary of the main modifications in installed infrastructure, only changing
optimum operational conditions of the CDU system for operational conditions and Crude Oil Total Flow.
two initial estimations are shown in Table 6.
Table 8. Optimization profits of the CDU system (US$/day)
The product yields are listed in Table 7. The initial
estimation 1 (IE-1) is a typical operational point of real
IE-1 Optimum EI-1 IE-2 Optimum EI-2
plant, and the initial estimation 2 (IE-2) is a converged
simulation point in PRO/II. The results of Tables 6-8 1.194.722 1.751.742 1.224.251 1.697.415
were obtained after applying the algorithm described in
Figure 2. Therefore, these constitute feasible optimum
points (with convergence in PRO/II). Figure 4 shows the ASTM-D86 distillation curves
found in GAMS and validated in PRO/II of some atmos-
Table 8 shows the system profit (Pr) before and after pheric products for the CDU system, later optimized by
optimization. The optimum point from EI-1 shows grea- IE-1. These results are evidence of good Metamodels
ter profit than the value found by IE-2, with an increase prediction within the optimization model.
1.600
1.400
1.200
1.000
Temperature (°F)
800
600
400
200
0
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
% Recuperated
Diesel ADT1 GAMS Diesel ADT2 GAMS Diesel ADT3 GAMS REDC ADT1 GAMS REDC ADT2 GAMS
REDC ADT3 GAMS JET ADT1 GAMS JET ADT2 GAMS JET ADT3 GAMS Diesel ADT1 PRO/II
Diesel ADT2 PRO/II Diesel ADT3 PRO/II Jet ADT1 PRO/II Jet ADT2 PRO/II Jet ADT3 PRO/II
REDC ADT1 PRO/II REDC ADT2 PRO/II REDC ADT3 PRO/II
● The profit increment of the system without investment Pinto, J.M., Joly, M. & Moro, LFL. (2000). Planning and
Scheduling Models for Refinery Operations. Comput.
and modifications in infrastructure contemplated in
Chem. Eng., 24: 2259-2276
this study, demonstrates the importance of optimiza-
tion models in industrial plants. However, optimization Wenkai, Li & Chi-Wai., Hui & AnXue, Li. (2005) Integrating
still has the feedstock fixed (constant composition) to CDU, FCC and product blending models into refinery
the units, an aspect that determines the plant operation. planning. Comput. Chem. Eng., 29 (9), 2010-2028.